5 Myths about Predictive Analytics
By Jamie MacLennan, CTO, Predixion Software
Modern predictive analytics tools and solutions are no longer restricted to use by IT specialists alone. They’re being used on business users’ desktops around the world, and rightly so.
In this article we will debunk five common myths that may have you thinking that predictive analyst is out of reach.
Myth #1: You can’t start until a data warehouse is in place
One reason many organizations don’t take advantage of predictive analytics is that they believe that a heavyweight data infrastructure must be in place to get started. They spend years trying to build and cost justify the existence of a data warehouse, including the component applications, and then miss the mark entirely, never realizing the benefits that predictive analytics provides.
Although a comprehensive data strategy is valuable, these organizations are missing countless opportunities to benefit from predictive analytics during the planning and implementation period. Predictive analytics is the “cherry” on top of a completed data warehouse, but not adopting predictive analytics because you don’t have a robust data warehouse is like not going skiing because you don’t own a lodge in Vail.
Modern predictive analytics tools don’t always need the perfect storm of a complete data infrastructure to provide meaningful and immediately applicable results. Today’s tools can handle a much wider variety of data situations. Many allow end users to glean insights into their own personal data while working with familiar tools -- such as Microsoft Excel. With self-service, easy-to-use predictive analytics tools, any and all information workers can leverage predictive analytics with the day-to-day data they have at their disposal.
Myth #2: Predictive analytics requires a Ph.D. or math degree
Many years ago, I was at a conference and met a leader in the predictive analytics market who said, “Regular business users aren’t capable of doing predictive analytics.” Speaking with many customers since then, it seems that this is a generally accepted belief. Organizations choose not to take advantage of predictive analytics because they believe they can’t afford -- or simply can’t find -- the human capital required to perform this highly specialized task.
The belief that only those with advanced mathematical degrees or statistical training are qualified and able to leverage predictive analytics is an artifact from the past. It’s similar to the long-standing argument that meaningful applications for computers could only be created by highly trained and specifically educated specialists which is now debunked (virtually anyone can develop meaningful Web or even iPhone applications). The complexity in implementing predictive analytics is a software problem, not a user problem.
Modern predictive analytics removes complexity and provides an environment where any data-savvy information worker can analyze data and retrieve meaningful, useful results without needing to understand how the mathematical algorithms operate in the background. Although having the best training and education in the field may yield improved results, the software itself can adequately present and dissect the data problems so that the business users who apply the tools will gain valuable and actionable insights that they simply wouldn’t get from traditional analytics.
Myth #3: There is a long time-to-value with predictive analytics
Traditional predictive analytics implementations have always worked like science experiments: data scientists perform some initial research, build hypotheses, and test those hypotheses by building predictive models. After iterative testing and model reevaluation, the “experiment” is put into “production.” Deploying a predictive model into an environment where it actually provides material business value requires specific application programming and changes in business practices. Each step of this process can easily take months of elapsed time.
Facing the perception of huge up-front software implementations, human capital requirements, and opportunity costs for a benefit that may not materialize for more than a year, many organizations simply feel that predictive algorithms are “not for us.” However, what if predictive analytics could identify right now the most important indicator to turn a lead into a paying customer? What if predictive analytics could tell you right now how to segment contacts to provide the best message for each group? What if predictive analytics could indicate right now the relationships in sales trends between different products? With modern predictive analytics tools, such results are possible.
Predictive analytics has evolved from a back-room activity (where results were embedded into the deepest level of an organization’s business processes) to a tool that knowledge workers can use every day to help them make better decisions about actions to take right now. Although traditional, large-scale predictive projects still provide great value and competitive advantage, modern predictive analytics software allows you to deploy predictive analytics to all your end users, offering them a better understanding of how data drives your business and helping them make better choices every day.
Myth #4: Curiosity killed the cat
Another reason people shy away from predictive analytics is the myth that somehow they can do something wrong. Purveyors of predictive technology have reinforced the idea that predictive analytics is an art, with an almost mystic quality about it, and that the mere attempt at performing such “magic” will bring failure and doom to the enterprise.
In reality, performing predictive analytics is putting an automated tool to work on data. The reason you execute such tasks is to gain information that will better assist you in making decisions that will improve your business. Predictive analytics provides you with that information -- at worst, you will get no useful information just as you would by running an irrelevant report -- but you won’t break anything. It is important to understand that you can use modern predictive analytic tools to extract the best information from your data rather than proscribe a set course of action.
Avoiding predictive analysis because you feel you may get something wrong is like avoiding GPS because it may give you bad directions. The GPS will get you there -- maybe not via the most direct route, but it will do a better job than you driving around randomly -- and if the GPS goofs and provides wrong directions, you are always in control of the wheel.
Myth #5: The results are incomprehensible
Yes, traditional predictive analytics tools often produced inscrutable results. Even if you could successfully navigate the toolsets to produce predictive models, results such as coefficients, p-values, and scores prevented new users from understanding the results and integrating new insights into business practices.
However, modern predictive analytic toolsets change that paradigm. Instead of presenting predictive results as a mashup of hyper-specialized statistics, modern predictive analytics demonstrates their findings in easy-to-consume visualizations that most information workers can understand. For example, simple bar charts are used to demonstrate which factors are the strongest in indicating customer churn or a patient’s likelihood to be readmitted to a hospital. Even the most tried-and-true measure returned from predictive analytics -- the score -- is simply an indication of the likelihood of a possible outcome occurring and can be understood with little training or relabeled for consumption by a larger audience.
A Final Word
Modern predictive analytics tools and solutions have moved from the domain of the quants in the back room of the IT departments to desktops across entire organizations. Don’t believe the myths that exist. There are great benefits available if you approach predictive analytics with a new mind-set.
Jamie MacLennan has been developing machine learning and analytic software for over a dozen years and is now the chief technology officer at Predixion Software. You can contact the author at firstname.lastname@example.org